反馈丢失条件下的即时译码网络编码研究
发布时间：2018-10-30 08:42
【摘要】：网络编码可显著提高无线网络系统性能,但是传统网络编码中的信宿节点需收到足够多的编码信息才能译码,不能实现实时译码。即时译码的网络编码技术(Instantly Decodable Network Coding,IDNC)通过选择相应数据包进行编码,确保信宿节点可实时译码,因此受到了当前学者的广泛关注。当前IDNC的研究大多基于理想的无线链路,但是现实环境中无线链路由于受到衰落、干扰等因素的影响,传输数据不可避免会出现差错。因此本论文将研究无线链路反馈丢失条件下的IDNC,主要做了如下几方面工作:1、在反馈丢失条件下,建立一种将IDNC和随机线性网络编码(Random Linear Network Coding,RLNC)相关联的网络编码模型。依据建立的IDNC图,提出最优IDNC解决方案下子代的概念,在每个子代中应用RLNC编码模型。研究表明,子代大小介于1和IDNCC之间时,系统性能介于IDNC和RLNC之间,且IDNC和RLNC只是该模型下具有特定子代大小的两个极端例子。2、在反馈丢失的多组播网络中,为了使IDNC的完成时延和译码时延最小化,提出一种最有可能状态找寻(Most Likely State Searching,MLSS)策略。通过MLSS方法建立多组播条件下的IDNC图,依据该模型得出平均完成时延和译码时延的表达式,采用最大权顶点找寻算法找出使两者分别最小化的团。仿真结果表明,相比于其他文献利用盲IDNC图更新法找寻最优编码包的算法,MLSS方法可获得更优的系统性能。3、为了权衡IDNC的完成时延和译码时延两种性能,提出一种多组播网络反馈丢失下的均衡完成时延和译码时延(Balance between Completion Time and Decoding Delay with Feedback Loss,BCD-FL)方案。首先用随机最短路径(Stochastic Shortest Path,SSP)分析IDNC在广播网络中的完成时延和译码时延均衡问题,利用分析出的性能表达式得到多组播网络下BCD-FL问题的SSP描述,从而在IDNC图中采用BCD-FL算法找出最大团。研究表明,提出的BCD-FL方案可较好均衡完成时延和译码时延性能。
[Abstract]:Network coding can significantly improve the performance of wireless network systems, but the traditional network coding nodes need to receive enough coding information in order to decode, can not achieve real-time decoding. The real-time decoding network coding technology (Instantly Decodable Network Coding,IDNC) ensures that the host node can be decoded in real time by selecting the corresponding packets to encode, so it has been widely concerned by scholars. Most of the current research on IDNC is based on the ideal wireless link, but in the real environment, due to the influence of fading, interference and other factors, the transmission data will inevitably appear errors. Therefore, this paper will study the IDNC, under the condition of wireless link feedback loss as follows: 1. Under the condition of feedback loss, we establish a kind of IDNC and random linear network coding (Random Linear Network Coding,. RLNC) the associated network coding model. According to the established IDNC diagram, the concept of offspring under the optimal IDNC solution is proposed, and the RLNC coding model is applied in each generation. The results show that when the offspring size is between 1 and IDNCC, the system performance is between IDNC and RLNC, and IDNC and RLNC are only two extreme examples with specific offspring size in this model. In order to minimize the completion delay and decoding delay of IDNC, a most probable state finding (Most Likely State Searching,MLSS) strategy is proposed. The IDNC graph under the condition of multi-multicast is established by MLSS method. According to the model, the expressions of average completion delay and decoding delay are obtained, and the max-weight vertex finding algorithm is used to find out the clusters which minimize the two groups respectively. The simulation results show that compared with other algorithms using blind IDNC graph updating method to find the optimal coding packet, the MLSS method can achieve better system performance. 3. In order to balance the completion delay and decoding delay of IDNC, the proposed method can achieve better performance. This paper presents a scheme for equalizing completion delay and decoding delay (Balance between Completion Time and Decoding Delay with Feedback Loss,BCD-FL under feedback loss in multi-multicast networks. Firstly, the completion delay and decoding delay equalization problem of IDNC in broadcast network is analyzed by stochastic shortest path (Stochastic Shortest Path,SSP), and the SSP description of BCD-FL problem in multi-multicast network is obtained by using the performance expression. Thus, the BCD-FL algorithm is used to find out the largest cluster in the IDNC diagram. The results show that the proposed BCD-FL scheme can equalize the performance of delay and decoding delay.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN92
本文编号：2299512
[Abstract]:Network coding can significantly improve the performance of wireless network systems, but the traditional network coding nodes need to receive enough coding information in order to decode, can not achieve real-time decoding. The real-time decoding network coding technology (Instantly Decodable Network Coding,IDNC) ensures that the host node can be decoded in real time by selecting the corresponding packets to encode, so it has been widely concerned by scholars. Most of the current research on IDNC is based on the ideal wireless link, but in the real environment, due to the influence of fading, interference and other factors, the transmission data will inevitably appear errors. Therefore, this paper will study the IDNC, under the condition of wireless link feedback loss as follows: 1. Under the condition of feedback loss, we establish a kind of IDNC and random linear network coding (Random Linear Network Coding,. RLNC) the associated network coding model. According to the established IDNC diagram, the concept of offspring under the optimal IDNC solution is proposed, and the RLNC coding model is applied in each generation. The results show that when the offspring size is between 1 and IDNCC, the system performance is between IDNC and RLNC, and IDNC and RLNC are only two extreme examples with specific offspring size in this model. In order to minimize the completion delay and decoding delay of IDNC, a most probable state finding (Most Likely State Searching,MLSS) strategy is proposed. The IDNC graph under the condition of multi-multicast is established by MLSS method. According to the model, the expressions of average completion delay and decoding delay are obtained, and the max-weight vertex finding algorithm is used to find out the clusters which minimize the two groups respectively. The simulation results show that compared with other algorithms using blind IDNC graph updating method to find the optimal coding packet, the MLSS method can achieve better system performance. 3. In order to balance the completion delay and decoding delay of IDNC, the proposed method can achieve better performance. This paper presents a scheme for equalizing completion delay and decoding delay (Balance between Completion Time and Decoding Delay with Feedback Loss,BCD-FL under feedback loss in multi-multicast networks. Firstly, the completion delay and decoding delay equalization problem of IDNC in broadcast network is analyzed by stochastic shortest path (Stochastic Shortest Path,SSP), and the SSP description of BCD-FL problem in multi-multicast network is obtained by using the performance expression. Thus, the BCD-FL algorithm is used to find out the largest cluster in the IDNC diagram. The results show that the proposed BCD-FL scheme can equalize the performance of delay and decoding delay.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN92
【参考文献】
相关期刊论文 前4条
1 王远;徐华;贾培发;;Design and analysis of a network coding algorithm for ad hoc networks[J];Journal of Central South University;2015年04期
2 戴彬;曹志刚;杨军;黄辰;王芙蓉;;基于匹配理论的无线广播编码重传算法[J];软件学报;2011年11期
3 许胤龙;詹成;罗文;李沛;;Ad hoc网络中基于网络编码的可靠组播[J];中国科学技术大学学报;2008年07期
4 刘祖云;胡进德;;分布式共享存储研究[J];成都大学学报(自然科学版);2008年01期
相关博士学位论文 前1条
1 唐斌;面向无线网络的高效网络编码方法研究[D];南京大学;2014年
相关硕士学位论文 前1条
1 杨叶舒;无线网络中网络编码子图和网络编码方法的研究[D];南京邮电大学;2014年
，本文编号：2299512
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2299512.html
